In News:

India is witnessing a significant leap in artificial intelligence innovation with the launch of two indigenous AI models — Kruti by Krutrim and BharatGPT Mini by CoRover. These developments aim to democratize AI access across the country by addressing local needs, multilingual capabilities, and infrastructure limitations.

Kruti: India’s First Agentic AI Assistant

Developed by Krutrim, the AI startup co-founded by Bhavish Aggarwal (of Ola fame), Kruti is positioned as India’s first agentic AI, going beyond conventional chatbots. Launched in 2025, Kruti integrates task execution capabilities such as:

• Cab booking
• Food ordering
• Bill payments
• Image generation
• Research assistance

Kruti is powered by Krutrim V2, a locally trained large language model (LLM), and combines open-source AI systems to deliver scalable, cost-effective, and contextualised solutions.

Key Features of Kruti

• Multilingual Support: Understands voice and text in 13 Indian languages
• Personalised AI: Learns user preferences, adapts tone and content
• Human-Centric Design: Supports read-aloud responses, summarised answers, stories, and tables
• SDK for Developers: Offers embeddable tools for LLM orchestration and task automation
• Integrated Assistant: Eliminates app-switching fatigue through contextual task handling

Aggarwal highlighted that Kruti is built for “how Indians live”—mobile-first, intuitive, and multilingual—offering free access to advanced AI tools.

Strategic Investment and Open AI Ecosystem

Krutrim has committed ?12,000 crore in investment (?2,000 crore already, ?10,000 crore by next year), launched Krutrim AI Lab, and published technical resources, with contributions to the open-source community. The company is positioning itself as a competitive force against global giants like OpenAI, Google, and Anthropic, and local firms like Sarvam AI.

BharatGPT Mini: Small Language Model for Bharat

On the same day, CoRover, a conversational AI firm, unveiled BharatGPT Mini, a small language model (SLM) with 534 million parameters trained on its proprietary conversational dataset.

• Supports 14 Indian languages
• Designed for low-compute, low-infrastructure environments
• Enables offline and edge deployments for fast, privacy-centric performance
• Ideal for underserved and rural regions with limited internet or device capacity

SLMs like BharatGPT Mini are emerging as viable tools for domain-specific, lightweight, and privacy-respecting AI in India, complementing the role of LLMs in more complex tasks.